Handling multi-language

[Melanie Courtot]

Hi,

Does anyone have good best practices/processes to handle multi-language resources?

Specifically we lead development of the Data Use Ontology (DUO), for which we are adding Japanese, French, German and Spanish translations.

Some questions:

- The main resource is authored in english, which means there can be a lag for other groups to produce the translation(s) when updates are made. Ideally, I'd like to release a merged file with all languages in, but I'm worried that if for example the French translation is delayed it would delay the release of everything. Or we could have releases without a specific translation which wouldn't be ideal for the users point of view. 

- I am planning to use dated version IRIs for translations files, which I store separately. I think this is the right way to go, but would welcome feedback.

- Do you have example of Makefile from the OBO github template framework that handles multi-language files?

Any other feedback would also be helpful! 

Thanks,

Melanie

-- 
Mélanie Courtot, PhD
Metadata standards coordinator
European Bioinformatics Institute (EMBL-EBI)

[James A. Overton]

Hi Melanie,

These are all important questions. I was hoping somebody would chime in with the perfect solution. :^)

We should have good answers for OBO, but unfortunately we don't yet. We're making progress with our tools, but we don't have policies and best practises yet.

I'm sure what you are proposing would work. Since you asked for advice, this is what I think I would do if it were my project:

1. use ROBOT templates for all my terms, with the logic and (primary) English annotations

2. have another ROBOT template for each language, and use the language support in ROBOT template to specify "AL rdfs:label@fr", etc.

3. merge all of these into one versioned release file, duo.owl

4. when there are enough changes, make a new versioned release of duo.owl

Then I would try to help my users extract the information they want from the main release file. ROBOT remove/filter can select by language tag, and SPARQL is another good option. If that just isn't possible, I guess I would create language-specific versioned OWL files as "views", as you proposed.

In my opinion, the biggest missing piece is multi-language support in ontology browsers. That's not something that OBO controls.

Whatever you choose. I'm happy to help you with your Makefile.

James

On Feb 17, 2021, at 09:52, Melanie Courtot <mcou...@gmail.com> wrote:



--
You received this message because you are subscribed to the Google Groups "obo-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to obo-discuss...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/obo-discuss/CALBDZOXkQapEaER1ZyMS-MsRUmwAmGqyvrEQgjr0_O1naOPt1A%40mail.gmail.com.

[John Graybeal]

I appreciated the detail in James' explanation, just one question about it: would each ROBOT template have some 'ontology cruft' that would be discarded on the merge, since there is only one overall ontology?  Or does ROBOT produce just the annotations in question, without ontology metadata?

I feel VERY keenly the absence of internationalization in BioPortal. I am super-eager to get it incorporated somehow. If anyone sees a good funding mechanism please let me know.

John

To view this discussion on the web visit https://groups.google.com/d/msgid/obo-discuss/EF73CEEF-1D0C-4237-A598-6055D725697A%40overton.ca.

========================

John Graybeal

Technical Program Manager

Center for Expanded Data Annotation and Retrieval /+/  BioPortal

[Balhoff, Jim]

Another big missing piece is support for language tags in OBO format. In my experience these are just lost. This format may or may not matter for different users.

[James A. Overton]

John: Yes, there is a bit of "ontology cruft", and yes, it will either be discarded by the merge or fixed with another ROBOT command such as `robot annotate`.

Jim: Good point about OBO-format. I didn't think of OBO-format because all my projects work from OWL and only convert to OBO-format at the end of the build process.

In theory OWL builds on the comprehensive language support in RDF (and XML). In practise, translating the content is the hard part, but it sounds like Melanie has that covered for DUO.

James

On Feb 24, 2021, at 11:59, John Graybeal <jgra...@stanford.edu> wrote:

 I appreciated the detail in James' explanation, just one question about it: would each ROBOT template have some 'ontology cruft' that would be discarded on the merge, since there is only one overall ontology?  Or does ROBOT produce just the annotations in question, without ontology metadata?

[Nico Matentzoglu]

I started an internationalisation channel in the OBO slack space to discuss further:

In case you have not joined yet:

https://join.slack.com/t/obo-communitygroup/shared_invite/zt-mp1ns8oi-A67RbfJJTVRczxkPLaJ37w

To view this discussion on the web visit https://groups.google.com/d/msgid/obo-discuss/7574BD11-7740-4DA4-A53E-6F2D733F5ED2%40overton.ca.